Different approaches to pseudo-R² naturally yield different results. For example, Nagelkerkes pseudo-R² tends to yield higher results than McFaddens pseudo-R².
As I am not a statistician, it thus can be somewhat difficult to interpret pseudo-R². In ecology (correct me if I am wrong), it is ± broadly accepted that a fair amount of the variability in the response variable is explained with McFaddens R² > 0.2, while a "good" Nagelkerkes R² is > 0.5.
When it comes to mixed models, I tried to use r.squaredGLMM {MuMIn}. However, I am quite unsure about the interpretation of the outcome. According to Why are results different between MuMIn::r.squaredGLMM and piecwiseSEM::sem.model.fits?, r.squaredGLMM implements Schielzeth and Nakagawa's R2 for generalized linear mixed effects models.
I suppose that both can vary between 0 and 1. Can anyone indicate how to interpret conditional and marginal pseudo-R² of r.squaredGLMM?
